| Muli County is located in the southeastern margin of the Qinghai-Tibet Plateau and the eastern side of the middle section of the Hengduan Mountains.It is a transitional zone between the Qinghai-Tibet Plateau and the Yunnan-Guizhou Plateau.The county is dominated by high-Zhongshan landform,forming a typical deep valley area of the western Sichuan Plateau.At the same time,the geological structure conditions are complex and the crustal movement is strong.Controlled by the influence of geological environment,the geological disasters in the area are highly developed.Landslide,collapse and frequent fire and debris flow have a great impact on the local development and life,so it is very necessary to carry out a systematic evaluation and research on the geological disasters in the area.In this paper,Muli County of western Sichuan Plateau is taken as a typical research area to deeply analyze the development characteristics and disaster causing mechanism of different types of geological disasters in the area.By summarizing the geological factors affecting the formation of disasters,reasonably selecting evaluation factors and grading them,and analyzing the data based on GIS technology,SPSS and MATLAB,The weighted information volume model(I-W),logistic regression information volume model(I-LR)and information volume and support vector machine coupling model(I-SVM)were respectively used to quantitatively evaluate the vulnerability of geological disasters in the study area.Meanwhile,the evaluation results and applicability of mathematical models were compared and analyzed,so as to master the vulnerability degree of geological disasters in the study area.To provide reference for local disaster prevention and mitigation work.The main achievements of this study are as follows:(1)The geological disasters in this area are mainly landslide and debris flow,followed by collapse.All kinds of disasters have the characteristics of regional dispersion and regional concentration.At the same time,they are greatly affected by human activities,and most of them are distributed around roads and along river valleys.Statistics showed that the average growth density was 3.07 spots/100km~2,0.14spots/100km~2and 0.88 spots/100km~2.(2)Geological disasters in the area are mainly controlled by topographic conditions,geological conditions,vegetation conditions,hydrological conditions and human engineering activities.They are mainly developed in the section with an elevation of 2000～3000m,slope of 20～30°,Ordovician and Triassic schist,SLATE,metamorphic sandstone,phyllitic rock strata,and vegetation coverage rate is within0.4～0.6.Disasters mainly occur around roads and along water systems.(3)Eight evaluation indexes including elevation,slope,slope direction,distance from water system,distance from road,distance from fault,NDVI and engineering geology rock group were selected and graded,and the information quantity value of each factor was calculated.The weight value of the evaluation index was calculated with the principal component analysis method,and the weighted information quantity model was constructed.With the information value as the independent variable and the occurrence of geological disasters as the dependent variable,the logistic regression calculation was carried out by using SPSS to obtain the regression coefficients of each factor,and the logistic regression information quantity model was constructed.The information value of the sample data is used to conduct training and testing in MATLAB.The radial basis kernel function(RBF)is selected according to the accuracy of the non-classifier model,and the optimal parameters of the model are calculated by genetic algorithm(GA)to construct the support vector machine model.The results of the three models showed similar spatial distribution characteristics.(4)The evaluation results of the three types of coupling models all showed the tendency of vulnerability from low to high,and the density of disaster points gradually increased.The ROC curve was used to test the accuracy of the model,and the AUC values were 0.901,0.923 and 0.918,respectively,indicating that the three models have high applicability and the logistic regression model has higher accuracy in the assessment of geological disaster susceptibility in this region. |
PDF Full Text Request